Evaluating NISAR's cropland mapping algorithm over the conterminous United States using Sentinel-1 data

Rose, Shannon; Kraatz, Simon; Kellndorfer, Josef; Cosh, Michael H.; Torbick, Nathan; Huang, Xiaodong; Siqueira, Paul

doi:10.1016/j.rse.2021.112472

Cited by 11 publications

(41 citation statements)

References 36 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…As in (Kraatz, Rose, et al., 2021; Kraatz, Torbick, et al., 2021; Rose et al., 2021), the performance metrics of overall accuracy (OA), Cohen’s Kappa or Youden’s J‐statistic (YJS) can be plotted as function of CV thr value, to show how they co‐vary and to those CV thr values yielding the best results—that is, the optimal CV thr value corresponding to max(OA), max(YJS) or max(Kappa). Cohen's Kappa is calculated at each pixel from TP, FP, FN, TN (Table 1) using Equations and , where Cohen's Kappa is (Cohen, 1960; McHugh, 2012):

\text{Kappa}=\left({p}_{0}-{p}_{e}\right)/\left(1-{p}_{e}\right)

{p}_{0}=(\text{TP}+\text{TN})/(\text{TP}+\text{FP}+\text{FN}+\text{TN})

with

\text{OA}={p}_{0}\ast 100,\ {p}_{e}={p}_{\gamma }-{p}_{N},

{p}_{\gamma }=(\text{TP}+\text{FP})\ast (\text{TP}+\text{FN})/{(\text{TP}+\text{FP}+\text{FN}+\text{TN})}^{2},

and

${p}_{N}=(\text{FN}+\text{TN})\ast (\text{FP}+\text{TN})/{(\text{TP}+\text{FP}+\text{FN}+\text...

…”

Section: Methodsmentioning

confidence: 93%

See 1 more Smart Citation

Evaluating the Robustness of NISAR's Cropland Product to Time of Observation, Observing Mode, and Dithering

Kraatz

Siqueira

Kellndorfer

et al. 2022

Earth and Space Science

View full text Add to dashboard Cite

Cropland mapping is important for monitoring agricultural practices, cropland distribution and for supporting food security programs. Radar remote sensing will likely provide a means of cropland mapping, which can be efficient, accurate, and globally applied. Algorithms used for spaceborne radar cropland mapping are still undergoing calibration and validation efforts, necessitating ground and aircraft‐based campaigns ahead of launch. The upcoming NASA ISRO SAR (NISAR) mission will map croplands at 1 ha spatial resolution using a coefficient of variation approach. NASA Jet Propulsion Laboratory recently released simulated NISAR data over cropland calibration sites including morning (AM), evening (PM), and dithered with and without gaps. These freely available simulated data allow for making accurate a priori estimates of how NISAR’s cropland science algorithm is impacted by factors such as orbit (i.e., the incidence angle, time of day), bandwidth (i.e., whether the 5, 20, or 40 MHz channels are used), and varying the system's pulse repetition frequency—a process known as dithering. This study investigates these science mission questions over agricultural regions in Mississippi and Arkansas, USA. Results show that compositing data from both orbits (AM + PM) yielded comparable classification accuracy to using only one of AM or PM data. The study also found that dithering only slightly decreased cropland mapping capability (<0.3% accuracy reduction).

show abstract

\text{Kappa}=\left({p}_{0}-{p}_{e}\right)/\left(1-{p}_{e}\right)

{p}_{0}=(\text{TP}+\text{TN})/(\text{TP}+\text{FP}+\text{FN}+\text{TN})

with

\text{OA}={p}_{0}\ast 100,\ {p}_{e}={p}_{\gamma }-{p}_{N},

{p}_{\gamma }=(\text{TP}+\text{FP})\ast (\text{TP}+\text{FN})/{(\text{TP}+\text{FP}+\text{FN}+\text{TN})}^{2},

and

${p}_{N}=(\text{FN}+\text{TN})\ast (\text{FP}+\text{TN})/{(\text{TP}+\text{FP}+\text{FN}+\text...

…”

Section: Methodsmentioning

confidence: 93%

“…The CV approach and performance assessment follows studies (Kraatz, Rose, et al., 2021; Kraatz, Torbick, et al., 2021; Rose et al., 2021). Specifically, the temporal CV is calculated as:

\text{CV}=\sigma /\mu

where σ and μ are the standard deviation and mean of the HV polarized RCS values over time for a given pixel.…”

Section: Methodsmentioning

confidence: 99%

Evaluating the Robustness of NISAR's Cropland Product to Time of Observation, Observing Mode, and Dithering

Kraatz

Siqueira

Kellndorfer

et al. 2022

Earth and Space Science

View full text Add to dashboard Cite

show abstract

“…Step three. Moreover, the coefficient of variation (CV) has been proven to be an effective indicator for distinguishing crop types and non-cropland (Huang et al, 2021;Liu et al, 2020;Rose et al, 2021;Whelen and Siqueira, 2018). The VH backscatter coefficient of crops, particularly of paddy rice, has a larger time-series variation range than non-agricultural land (e.g., urban and water) (Figure S3).…”

Section: Algorithm For Identifying Paddy Rice Fieldsmentioning

confidence: 99%

“…Based on this time-series characteristic, paddy rice may be identified from different land surfaces. Therefore, we removed pixels with CV values greater than 0.3 and less than 0.7 calculated using VH during the growth of paddy rice (Rose et al, 2021;Whelen and Siqueira, 2018). We measured the using the temporal mean (MEAN) and standard deviation (SD) of the time-series of VH during the paddy rice growth period:…”

Section: Algorithm For Identifying Paddy Rice Fieldsmentioning

confidence: 99%

AsiaRiceMap10m: high-resolution annual paddy rice maps for Southeast and Northeast Asia from 2017 to 2019

Han

Zhang

Luo

et al. 2021

Preprint

View full text Add to dashboard Cite

Abstract. An accurate paddy rice map is crucial for food security, particularly for Southeast and Northeast Asia. The MODIS satellite data is useful for mapping paddy rice at continental scales but has a problem of mixed-pixel due to coarse spatial resolution. To reduce the mixed pixels, we designed a rule-based method for mapping paddy rice by integrating time-series Sentinel-1 and MODIS data. We demonstrated the method in generating annual paddy rice maps for Southeast and Northeast Asia in 2017–2019 (AsiaRiceMap10m). The resultant paddy rice maps were compared with available agricultural statistics at subnational levels and existing rice maps for some countries. The results show that the linear coefficient of determination (R2) between our paddy rice maps and agricultural statistics ranges from 0.80 to 0.97. The paddy rice planting areas in 2017 were spatially consistent with the existing maps in Vietnam (R2 = 0.93) and Northeast China (R2 = 0.99). The spatial distribution of the 2017–2019 composite paddy rice map is consistent with the rice map from the International Rice Research Institute. The paddy rice planting area may be underestimated in the region where the flooding signal is not strong. The dataset is useful for water resource management, rice growth, and yield monitoring. The full product is publicly available at https://doi.org/10.17632/j34b3jsvr9.1 (Han et al., 2021a). Find small examples here (https://doi.org/10.17632/cnc3tkbwcm.1) (Han et al., 2021b).

show abstract

“…Based on this time-series characteristic, paddy rice may be identified from different land surfaces. Therefore, we removed pixels with CV values greater than 0.3 and less than 0.7 calculated using VH during the growth of paddy rice (Rose et al, 2021;Whelen and Siqueira, 2018). We measured the 𝐶𝑉 𝑉𝐻 using the temporal mean (MEAN) and standard deviation (SD) of the time-series of VH during the paddy rice growth period:…”

Section: Algorithm For Identifying Paddy Rice Fieldsmentioning

confidence: 99%

Reply on RC2

zhang

2021

View full text Add to dashboard Cite

Evaluating NISAR's cropland mapping algorithm over the conterminous United States using Sentinel-1 data

Cited by 11 publications

References 36 publications

Evaluating the Robustness of NISAR's Cropland Product to Time of Observation, Observing Mode, and Dithering

Evaluating the Robustness of NISAR's Cropland Product to Time of Observation, Observing Mode, and Dithering

AsiaRiceMap10m: high-resolution annual paddy rice maps for Southeast and Northeast Asia from 2017 to 2019

Reply on RC2

Contact Info

Product

Resources

About